Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2372041681
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ohashi, K.
Right arrow Articles by Berbaum, K. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ohashi, K.
Right arrow Articles by Berbaum, K. S.
(Radiology 2005;237:570-577.)
© RSNA, 2005

Musculoskeletal Imaging

Orthopedic Hardware Complications Diagnosed with Multi–Detector Row CT1

Kenjirou Ohashi, MD, Georges Y. El-Khoury, MD, D. Lee Bennett, MD, José M. Restrepo, MD and Kevin S. Berbaum, PhD

1 From the Department of Radiology, University of Iowa Roy J. and Lucille A. Carver College of Medicine, 200 Hawkins Dr, Iowa City, IA 52242. From the 2003 RSNA Annual Meeting. Received September 30, 2004; revision requested December 7; revision received January 20, 2005; accepted February 21. Address correspondence to K.O. (e-mail: kenjirou-ohashi{at}uiowa.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
PURPOSE: To retrospectively evaluate multi–detector row computed tomography (CT) for the depiction of orthopedic hardware complications in the spine and appendicular skeleton.

MATERIALS AND METHODS: This HIPAA-compliant study had institutional review board approval; patient informed consent was not required. Results of 114 multi–detector row CT studies performed because of possible hardware complications in 109 patients (57 men, 52 women; mean age, 44 years; age range, 12–82 years) were available for analysis. The CT studies were retrospectively reviewed and compared with clinical or surgical outcomes, which were used as the reference standard. In another experiment, detection of hardware complications on radiographs and multi–detector row CT images was compared between two readers for selected cases (18 positive and 26 negative) by using receiver operating characteristic (ROC) methods.

RESULTS: For 91 (80%) of 114 multi–detector row CT studies, the complication status could be determined on the basis of clinical or surgical outcomes. Twenty-three multi–detector row CT studies were confirmed to be positive (revealing 10 cases of nonunion, five cases of hardware malplacement, three cases of hardware loosening, three perihardware fractures, and two chronic infections), and 57 were confirmed to be negative. There were three false-positive and eight false-negative multi–detector row CT studies. With clinical or surgical outcomes as the reference standard, the sensitivity, specificity, and positive and negative predictive values of multi–detector row CT were 74% (23 of 31 studies), 95% (57 of 60 studies), 88% (23 of 26 studies), and 88% (57 of 65 studies), respectively. Results of ROC analysis indicated that detection of hardware complications was much lower with radiography than with multi–detector row CT (area under ROC curve, 0.84 vs 1.00; F = 4.69, df = 1, 43; P < .05).

CONCLUSION: Multi–detector row CT is an effective tool for depicting orthopedic hardware complications.

© RSNA, 2005


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
In the past, metal artifacts from orthopedic hardware (metallic devices used for orthopedic fixation or arthroplasty) at computed tomography (CT) that interfere with the visualization of nearby soft tissues and bones have prevented the routine use of CT in patients with orthopedic hardware (1,2). The advent of multi–detector row CT, combined with advancements in computer workstations (in terms of both hardware and software), has resulted in a marked reduction in the severity of metal artifacts (3–6). These new technologies have not only reduced the severity of metal artifacts but have also enhanced our ability to evaluate abnormalities related to hardware located within complex anatomic structures. Until recently, these abnormalities were evaluated primarily with conventional radiography.

At our institution, we use radiography followed by multi–detector row CT when hardware complications are suspected. To our knowledge, no formal study has evaluated the performance of multi–detector row CT in the detection of orthopedic hardware complications. Nuclear medicine studies such as indium 111–labeled leukocyte scintigraphy may be ordered specifically to rule out infection (7–9). However, in the majority of patients, anatomic details of the bone and soft tissue need to be evaluated initially. Magnetic resonance imaging has been used for the evaluation of periprosthetic infection (10,11) but is not indicated for the evaluation of more common complications such as nonunion and hardware malplacement. Thus, the purpose of our study was to retrospectively evaluate multi–detector row CT for the depiction of orthopedic hardware complications in the spine and appendicular skeleton.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Patients and Reference Standard
Our institutional review board approved this retrospective study; patient informed consent was not required. The study was conducted in accordance with the federal Health Insurance Portability and Accountability Act. We included consecutive patients who were suspected of having orthopedic hardware complications and were examined with multi–detector row CT during a 7-month period (from October 1, 2002, to April 30, 2003). We excluded 11 patients who had degenerative joint disease or synostosis and 12 patients whose follow-up information was not available. As a result, 114 multi–detector row CT studies performed in 109 patients (57 men, 52 women; mean age, 44 years; age range, 12–82 years) were available for review and analysis. Five patients had undergone multi–detector row CT at two anatomic locations. For 54 (47%) of the 114 studies, multi–detector row CT had been performed within 30 days after surgery (acute cases; mean interval, 3 days; range, 0–16 days) and for 60 (53%) of the studies, multi–detector row CT had been performed 30 days or more after surgery (chronic cases; mean interval, 2 years 1 month; range, 1 month to 19 years 5 months).

To validate the diagnoses at multi–detector row CT, surgical outcomes (available for 49 of the 114 studies) or clinical outcomes (available for 65 studies) were used as the reference standard. These were reviewed by a musculoskeletal radiologist (K.O.). Clinical criteria for excluding hardware complications included an improvement in symptoms without surgical intervention or localized symptoms that could not be explained by the hardware location at follow-up. Patients who had undergone uneventful two-stage surgery (external and internal fixation) were considered not to have a complication when they were asymptomatic at follow-up. The surgical outcome criteria for hardware complications included an improvement in symptoms after surgical intervention (hardware revision or removal). Detection of microorganisms in cultures or the presence of numerous polymorphonuclear leukocytes in pathologic specimens (12) was considered to be evidence of infection. The condition of patients who had persistent symptoms around the hardware site whether or not surgical intervention (hardware revision or removal) had been performed, as well as the condition of patients who had undergone additional nonrevision surgery in the region of the original hardware, was classified as indeterminate for hardware complications. Thus, only patients who had undergone surgical intervention (hardware revision or removal) with a favorable outcome were considered to have had complications.

Imaging
Radiographic studies (106 studies in 101 patients) were performed by using either a conventional screen-film radiography system with a 150-kV 800-mA x-ray generator (KXO-80F; Toshiba American Medical Systems, Tustin, Calif) or a computed radiography system (CR 950; Eastman Kodak Company, Rochester, NY). At least two views (range, two to four views; mean, 2.6 views) were obtained for each anatomic site for the observer performance study cases (see below). Standard views included anteroposterior and lateral (with or without flexion and extension) views of the cervical, thoracic, and lumbar spine; anteroposterior and both oblique views of the pelvis; and anteroposterior and lateral (with or without oblique) views of the extremities.

CT was performed by using a four–detector row helical CT scanner (Aquilion; Toshiba American Medical Systems) and the following parameters: 135 kVp, a 0.5-second scanning time per gantry rotation, and a 512 x 512 matrix. Other parameters, including tube current, field of view, thickness, table travel per rotation, reconstruction thickness, and overlap for each anatomic location are summarized in Table 1. We used the standard soft-tissue kernel. The reconstructed images were sent to a computer workstation (Vitrea 2, version 3.3; Vital Images, Plymouth, Minn) over an intradepartmental picture archiving and communication system network (Eastman Kodak Company) by using the Digital Imaging and Communications in Medicine protocol. Multiplanar reformatted images and other three-dimensional (3D) images were reviewed on the computer workstation. Multiplanar reformatted images were reviewed by using window widths of 1700–9000 and window levels of 800–2000.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Multi–Detector Row CT Acquisition Parameters

 
Image Interpretation
A musculoskeletal radiologist (K.O., with 5 years of experience in musculoskeletal multi–detector row CT) retrospectively reviewed the 114 multi–detector row CT studies, initially while blinded to the patient's clinical history, and then compared the CT findings with medical records, including radiology reports. If the interpretation in the original radiology report and the retrospective interpretation were discrepant, two other musculoskeletal radiologists (G.Y.E. and D.L.B., both with 6 years of experience in musculoskeletal multi–detector row CT) independently evaluated these CT studies further while blinded to the patient's clinical history and the other reader's interpretation. This yielded two additional retrospective interpretations. The most concordant readings among the three retrospective interpretations constituted the final interpretation. Clinical data—including surgical procedures, indications for surgery, and types of orthopedic hardware—were recorded.

Each hardware complication was categorized as representing one of the following: hardware malplacement, loosening and/or dislodgment, nonunion or nonfusion with or without hardware fracture, perihardware fracture, or "other." The imaging criteria for nonunion or nonfusion included a lack of osseous bridging across the fracture or across the joint, respectively. An area of perihardware lucency greater than 2 mm in thickness was considered to indicate loosening. Malplacement of hardware was considered to be present when an abnormal hardware position with hardware impingement on an important structure such as a nerve or vessel was seen. Penetration of hardware into any joint except the facet joint was considered to represent malplacement of the hardware. Periosteal reactions, areas of focal lucency, sequestra, areas of bone sclerosis, and associated soft-tissue masses were considered signs of infection. Complications of loosening and/or dislodgment and nonunion or nonfusion were primarily evaluated, regardless of coexisting findings of infection.

Observer Performance Study
In the second part of our investigation, two musculoskeletal radiologists (G.Y.E. and D.L.B.) independently reviewed studies obtained in selected patients who had undergone both CT and radiography for hardware evaluation. The images were evaluated by using printed hard copies for radiographic interpretation and a computer workstation for multi–detector row CT interpretation. This observer performance study was conducted 3 months after the independent case review of the CT studies to mitigate recall bias. Eighteen patients with clinically proved hardware complications and 26 patients with a clinically proved absence of hardware complications were included as the sample for the receiver operating characteristic (ROC) observer performance study. These patients were selected as follows. For each of the patients with complications, a patient without a complication was selected to match in terms of hardware anatomic location and chronicity (length of time the hardware had been in the patient); this resulted in 18 paired cases. The matching was limited owing to a lack of radiographic studies in eight patients (106 radiographic studies, vs 114 multi–detector row CT studies, had been performed). To improve the matching of studies showing normal findings with studies showing abnormal findings and to increase the sample size, studies in eight patients without hardware complications were added (to the 18 paired studies) during the period after our retrospective review. This Health Insurance Portability and Accountability Act–compliant addition had institutional review board approval, and the need for informed consent was waived.

The readers were blinded to the name, sex, age, and clinical outcome of the patients. The radiographic and multi–detector row CT studies were reviewed in a random order. For radiographic readings, old comparison radiographs were reviewed when available. Each reader's certainty that a complication was present was rated as a percentage (0%–100%) by using a continuous subjective probability response scale (13,14).

Statistical Analysis
Results of our retrospective review are largely descriptive. The sensitivity, specificity, and positive and negative predictive values of multi–detector row CT in the diagnosis of orthopedic hardware complications were calculated by using surgical or clinical outcomes as the reference standard. For the sample used for ROC analysis, the z test for the difference between population proportions was used to test whether the male-to-female ratio in the group with complications was different from that in the group without complications. Two-group t tests were used to test for differences in age, number of days between surgery and multi–detector row CT, and number of days between radiography and CT between the group with and the group without complications. The t tests were performed by using a software program (Excel 2000 [9.0.4402 SR-1]; Microsoft, Redmond, Wash).

The data from the observer performance study were analyzed as follows. A total of 88 observations (18 positive cases, 26 negative cases, and two readers) were analyzed by using ROC analysis. ROC curves from the ratings were fitted with the standard binormal model (15) by using a computer program (16). The area under the ROC curve (Az) was used to index detection accuracy. The multireader ROC method of Dorfman, Berbaum, Metz, and others (17,18) was used for statistical analysis. This method treats readers as a fixed factor; hence, our conclusions generalize to the population of patients (18). Individual ROC curves were also generated for the two readers for each modality. Because neither reader produced more than 20 rating categories with the continuous rating scales, all of the ROC rating data could be fitted by using standard methods (15,16).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Clinical Data
The multi–detector row CT studies consisted of 36 spine studies (four cervical, seven thoracic, and 25 lumbar), 35 pelvis and/or femur studies, 32 ankle and/or foot studies, five knee studies, three shoulder studies, and three elbow studies. Surgical procedures, indications, and types of orthopedic hardware are summarized in Table 2.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Initial Surgical Procedures for Orthopedic Hardware Placement in 114 Cases

 
Retrospective Multi–Detector Row CT Readings
For three multi–detector row CT studies, the retrospective interpretations were different from the original interpretations. The two additional independent readers agreed with the initial observer for all three studies. For two instances in which multi–detector row CT studies were originally regarded as indicating a partial fusion, the retrospective interpretation was nonfusion. Clinically, these two instances were treated surgically, and patient symptoms improved. One study was retrospectively interpreted as showing screw penetration into a joint; the original report did not mention this finding. The patient had no related symptoms at clinical follow-up.

Multi–Detector Row CT Findings versus Surgical and Clinical Outcomes
For each multi–detector row CT case, we used surgical outcomes or clinical follow-up as the reference standard. The term case refers to an individual multi–detector row CT study; because more than one anatomic area was examined with multi–detector row CT in five patients, the term case does not necessarily correspond to a single patient. The mean clinical follow-up period, as defined by the date of the last clinical note, was 5.4 months after the multi–detector row CT study.

Multi–detector row CT studies showed no evidence of hardware complications in 77 cases. For 65 of these cases, the complication status could be determined on the basis of surgical outcomes or findings at clinical follow-up. Among these 65 cases, 57 (88%) were considered to be true-negative and eight (12%) were considered to be false-negative. For the latter eight cases, the hardware was surgically removed, and symptoms resolved. For the remaining 12 cases, in which multi–detector row CT results were negative for hardware complications, the clinical or surgical outcomes were indeterminate for hardware complications.

Multi–detector row CT studies revealed hardware complications in 37 cases. For 26 of these cases, the complication status was proved with follow-up. Among these 26 cases, 23 (88%) were true-positive. The affected regions and complications in these 23 cases are summarized in Table 3. There were 10 cases of nonunion or nonfusion, with or without hardware fracture (Fig 1); three cases of hardware loosening and/or dislodgment, with or without infection (Fig 2); five cases of hardware malplacement (Fig 3); three cases of perihardware fracture; and two cases of chronic infection. All 23 cases were treated with additional surgery, and patient symptoms improved. Three cases in which multi–detector row CT indicated hardware complications were treated conservatively and symptoms improved; therefore, these three cases were considered to be false-positive on the basis of results of clinical follow-up. Multi–detector row CT showed malplacement of hardware in all three of these cases. For the remaining 11 cases, complication status could not be clinically determined because of the absence of clinical improvement after surgery (two cases) or a lack of surgical intervention and persistent symptoms (nine cases). Seven of these cases were associated with spinal hardware.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Twenty-three True-Positive Orthopedic Hardware Complications at Multi–Detector Row CT

 


View larger version (87K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1a. Nonunion of clavicular fracture in 37-year-old woman. (a) Anteroposterior radiograph of left clavicle shows plate and screw fixation for midclavicular fracture. (b) Coronal oblique CT image reconstructed along the long axis of the clavicle shows a bone gap (arrow) with sclerosis at the fracture site. (c) Transverse oblique CT image reconstructed parallel to the long axis of the clavicle shows similar findings (arrow) consistent with nonunion. (d) Volume-rendered CT image shows intact hardware.

 


View larger version (41K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1b. Nonunion of clavicular fracture in 37-year-old woman. (a) Anteroposterior radiograph of left clavicle shows plate and screw fixation for midclavicular fracture. (b) Coronal oblique CT image reconstructed along the long axis of the clavicle shows a bone gap (arrow) with sclerosis at the fracture site. (c) Transverse oblique CT image reconstructed parallel to the long axis of the clavicle shows similar findings (arrow) consistent with nonunion. (d) Volume-rendered CT image shows intact hardware.

 


View larger version (62K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1c. Nonunion of clavicular fracture in 37-year-old woman. (a) Anteroposterior radiograph of left clavicle shows plate and screw fixation for midclavicular fracture. (b) Coronal oblique CT image reconstructed along the long axis of the clavicle shows a bone gap (arrow) with sclerosis at the fracture site. (c) Transverse oblique CT image reconstructed parallel to the long axis of the clavicle shows similar findings (arrow) consistent with nonunion. (d) Volume-rendered CT image shows intact hardware.

 


View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1d. Nonunion of clavicular fracture in 37-year-old woman. (a) Anteroposterior radiograph of left clavicle shows plate and screw fixation for midclavicular fracture. (b) Coronal oblique CT image reconstructed along the long axis of the clavicle shows a bone gap (arrow) with sclerosis at the fracture site. (c) Transverse oblique CT image reconstructed parallel to the long axis of the clavicle shows similar findings (arrow) consistent with nonunion. (d) Volume-rendered CT image shows intact hardware.

 


View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a. Loosening of screws after spinal fixation in 35-year-old woman. (a) Lateral radiograph of lumbar spine shows posterior fusion hardware at L4, L5, and S1, with a cage at the L4-5 disk space. Perihardware lucency is not seen. Moderate anterior displacement of L5 on S1 is noted. (b, c) Sagittally reconstructed CT images through each pedicle show prominent osteolysis (arrows) around the S1 transpedicular screws.

 


View larger version (99K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b. Loosening of screws after spinal fixation in 35-year-old woman. (a) Lateral radiograph of lumbar spine shows posterior fusion hardware at L4, L5, and S1, with a cage at the L4-5 disk space. Perihardware lucency is not seen. Moderate anterior displacement of L5 on S1 is noted. (b, c) Sagittally reconstructed CT images through each pedicle show prominent osteolysis (arrows) around the S1 transpedicular screws.

 


View larger version (110K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2c. Loosening of screws after spinal fixation in 35-year-old woman. (a) Lateral radiograph of lumbar spine shows posterior fusion hardware at L4, L5, and S1, with a cage at the L4-5 disk space. Perihardware lucency is not seen. Moderate anterior displacement of L5 on S1 is noted. (b, c) Sagittally reconstructed CT images through each pedicle show prominent osteolysis (arrows) around the S1 transpedicular screws.

 


View larger version (142K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3a. Screw malplacement in 74-year-old woman who underwent posterior lumbosacral fusion. (a) Anteroposterior and (b) lateral views of the lumbosacral spine show posterior fusion hardware extending to the ilium. (c) Sagittal, (d) transverse, and (e) coronal images show right S1 screw tip (short arrow) in the neural foramen with impingement on the S1 nerve root (long arrow).

 


View larger version (124K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3b. Screw malplacement in 74-year-old woman who underwent posterior lumbosacral fusion. (a) Anteroposterior and (b) lateral views of the lumbosacral spine show posterior fusion hardware extending to the ilium. (c) Sagittal, (d) transverse, and (e) coronal images show right S1 screw tip (short arrow) in the neural foramen with impingement on the S1 nerve root (long arrow).

 


View larger version (104K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3c. Screw malplacement in 74-year-old woman who underwent posterior lumbosacral fusion. (a) Anteroposterior and (b) lateral views of the lumbosacral spine show posterior fusion hardware extending to the ilium. (c) Sagittal, (d) transverse, and (e) coronal images show right S1 screw tip (short arrow) in the neural foramen with impingement on the S1 nerve root (long arrow).

 


View larger version (98K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3d. Screw malplacement in 74-year-old woman who underwent posterior lumbosacral fusion. (a) Anteroposterior and (b) lateral views of the lumbosacral spine show posterior fusion hardware extending to the ilium. (c) Sagittal, (d) transverse, and (e) coronal images show right S1 screw tip (short arrow) in the neural foramen with impingement on the S1 nerve root (long arrow).

 


View larger version (118K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3e. Screw malplacement in 74-year-old woman who underwent posterior lumbosacral fusion. (a) Anteroposterior and (b) lateral views of the lumbosacral spine show posterior fusion hardware extending to the ilium. (c) Sagittal, (d) transverse, and (e) coronal images show right S1 screw tip (short arrow) in the neural foramen with impingement on the S1 nerve root (long arrow).

 
In summary, 91 of 114 multi–detector row CT studies could be evaluated by using surgical or clinical outcomes as the reference standard. There were 57 true-negative, 23 true-positive, eight false-negative, and three false-positive cases. All positive results were confirmed with results of additional surgical treatment and clinical follow-up. Our results yielded a sensitivity of 74% (23 of 31 cases), a specificity of 95% (57 of 60 cases), a positive predictive value of 88% (23 of 26 cases), and a negative predictive value of 88% (57 of 65 cases) for multi–detector row CT evaluation of patients clinically suspected of having hardware complications.

Observer Performance Study
Clinical data for the selected cases are summarized in Table 4. No significant difference was noted between the cases with and those without complications in terms of age, male-to-female ratio, average number of days between surgery and multi–detector row CT, and average number of days between radiography and multi–detector row CT. In 24 (55%) of the 44 cases, radiography had been performed after or on the same day as multi–detector row CT (mean, 17 days; range, 1–60 days). Radiography had been performed before multi–detector row in 20 (45%) cases (mean, 11 days; range, 0–65 days). In 39 (89%) cases, the radiographic and multi–detector row CT studies were performed within a 1-month period. Previous radiographs were available for comparison in 40 (91%) cases (mean interval between acquisition of current and acquisition of previous radiographs, 63 days).


View this table:
[in this window]
[in a new window]

  		TABLE 4. Clinical Data for Observer Performance Study Cases

 
When the Dorfman-Berbaum-Metz method was applied, the difference in Az between radiography and multi–detector row CT was found to be significant (0.84 vs 1.00; F = 4.69, df = 1, 43; P < .05). Figure 4 shows binormal ROC curves fitted to the data of reader 1, and Figure 5 shows binormal ROC curves fitted to the data of reader 2. These results indicate that the performance of radiography in the detection of hardware complications is much lower than that of multi–detector row CT.



View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4. Binormal ROC curves fitted to the data of reader 1. The Az for radiography was 0.87, whereas the Az for multi–detector row CT (MD-CT) was 1.00. P(FP) = probability of false-positive results, P(TP) = probability of true-positive results.

 


View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Binormal ROC curves fitted to the data of reader 2. The Az for radiography was 0.81, whereas the Az for multi–detector row CT (MD-CT) was 1.00. P(FP) = probability of false-positive results, P(TP) = probability of true-positive results.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
Modern metal alloys used in some kinds of orthopedic hardware produce fewer artifacts at CT. Prostheses made of a cobalt-chrome alloy cause substantial artifacts, whereas those made of titanium cause minimal artifacts (19). Steel alloys are widely used in fixation hardware, which constitutes most of the hardware used by our surgeons. One of the advantages of multi–detector row CT is its ability to reduce the severity of metal or streak artifacts (3), which are caused by photopenic holes in the x-ray projection (20). With multi–detector row CT, adjacent channels pick up excess tissue irradiation in the penumbra of the beam, which partially fills the photopenic holes in the projection. This is one of the rare instances in which scattered radiation is useful. Other factors that contribute to the reduction in the severity of metal artifacts include the use of increased peak kilovoltage and tube current settings, as well as the use of decreased collimation and pitch values. The use of such postprocessing techniques as a soft-tissue kernel, multiplanar reformation, and 3D reformation with shaded-surface display or volume rendering also contributes to producing CT images with less severe metal artifacts. An extended CT scale is used to optimize visualization of perihardware soft tissues (3–6,21–25).

Combined with advancements in high-speed postprocessing workstations and 3D imaging software, multi–detector row CT has brought multiplanar and 3D interpretation into daily practice by yielding isotropic or near-isotropic data sets. Rapid computer display of these isotropic or near-isotropic data sets as images in any arbitrary plane and as 3D images helps in evaluating complex anatomic structures such as the spine, pelvis, and joints. Transitioning from hard-copy image interpretation to the interpretation of multiplanar and 3D images that are rapidly displayed on a computer monitor has given radiologists substantial versatility in evaluating hardware in different anatomic locations. Fractures and joints are evaluated without much difficulty, regardless of hardware orientation or the orientation of the interface between the hardware and bone. Comparison of sequential radiographs may be helpful in detecting loosening or displacement of a prosthesis (26,27) and for evaluation of the progression of arthrodesis or fracture union. However, radiography is not always optimal for visualizing hardware and bone because of the variety of the geometric orientations of hardware and overlapping bone structures (28,29). Multiplanar reformations of multi–detector row CT data can reveal hardware placement and associated bone changes without obscuration. Three-dimensional images such as volume-rendered images may help in the detection of major hardware failures (eg, hardware fracture).

Diagnosis of a hardware complication is not always straightforward. Painful hardware is an example: Hardware, which is prominent under the skin, can be painful; this may be the only indication for removal even if there is no apparent complication at imaging (30,31). The prominence of hardware is thought to cause chronic tissue irritation that leads to pain (31). In our series, eight patients (all of whom had false-negative CT findings) were considered by the surgeons to have painful hardware because of the lack of apparent hardware complication at radiography or CT; this factor resulted in a decreased sensitivity for multi–detector row CT (74%). In the clinical setting, when hardware complications are suspected, the positive predictive value of multi–detector row CT (88% in our study) is more meaningful. On the other hand, patients may not be symptomatic in the setting of hardware failure or malplacement (31). Fractured fixation hardware may not be problematic if bone fusion has been completed. Therefore, imaging findings need to be correlated with symptoms and with findings at physical examination. In our institution, for most patients suspected of having hardware complications, we currently rely on multi–detector row CT findings in the detection of hardware failure and consideration of treatment options. At present, to our knowledge, no other formal reports verify the use of multi–detector row CT in the evaluation of orthopedic hardware complications.

There were certain limitations in this study. Because our study was performed at a tertiary medical center, the study population and types of surgical procedures were heterogeneous; however, the study did include most body parts, with spine, pelvis, and extremity studies each accounting for about one-third of the sample. The study included major surgical procedures, including fracture fixation, spinal fusion, and arthroplasty, although the total number of arthroplasties was limited. Because a wide variety of hardware placement techniques were used at various locations, further breakdown of our study population by anatomic location and/or surgical technique would not appear to be either practical or fruitful. Because this was a retrospective review, we did not standardize imaging techniques, and no detailed information on the metal materials used was available. The clinically suspected differential diagnosis before multi–detector row CT was not studied; therefore, no direct effect of the multi–detector row CT findings on clinical decision making was tested.

There were some additional limitations of our study design. The results of our observer study suggested that radiography is less accurate than multi–detector row CT for the diagnosis of hardware complications. The fact that a statistically significant difference could be demonstrated with only 44 studies and two readers suggests a large difference. However, it should be noted that the examination of patients with multi–detector row CT may result in the interpretation of a greater-than-normal proportion of radiographic studies as equivocal. On the other hand, the studies used in our ROC investigation were limited to those whose findings could be proved surgically or clinically. Although surgical treatments were performed in about two-thirds of the patients in whom a hardware complication was suspected at multi–detector row CT, this might have eliminated some of the most difficult CT and radiographic studies to interpret because surgery may not have been performed on some patients in whom the results of imaging studies were equivocal. Hence, the observed performance of both multi–detector row CT and radiography might be better than the true performance of these modalities, and the observed difference in performance between radiography and multi–detector row CT may be larger than the true difference. Nevertheless, our study results demonstrate that multi–detector row CT performs much better than radiography in the identification of clinically relevant hardware complications in patients in whom the diagnosis can be verified with favorable surgical outcomes.

Finally, our results provide evidence that multi–detector row CT, coupled with advances in computer hardware and software, is an effective imaging tool for the evaluation of orthopedic hardware. These results give scientific support for orthopedic surgeons to order a CT examination in patients suspected of having hardware complications. This is important because CT was previously avoided for patients with metallic hardware.


   FOOTNOTES
 

Abbreviations: Az = area under ROC curve • ROC = receiver operating characteristic • 3D = three-dimensional

Authors stated no financial relationship to disclose.

Author contributions: Guarantors of integrity of entire study, K.O., G.Y.E.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, K.O., K.S.B.; clinical studies, K.O., G.Y.E.; statistical analysis, K.O., J.M.R., K.S.B.; and manuscript editing, K.O., G.Y.E., D.L.B., K.S.B.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Young SW, Muller HH, Marshall WH. Computed tomography: beam hardening and environmental density artifact. Radiology 1983;148:279–283.[Abstract/Free Full Text]
  2. Golimbu C, Firooznia H, Rafii M, Engler G, Delman A. Computed tomography of thoracic and lumbar spine fractures that have been treated with Harrington instrumentation. Radiology 1984;151:731–733.[Abstract/Free Full Text]
  3. Rydberg J, Buckwalter KA, Caldemeyer KS, et al. Multisection CT: scanning techniques and clinical applications. RadioGraphics 2000;20:1787–1806.[Abstract/Free Full Text]
  4. Fishman EK, Magid D, Robertson DD, Brooker AF, Weiss P, Siegelman SS. Metallic hip implants: CT with multiplanar reconstruction. Radiology 1986;160:675–681.[Abstract/Free Full Text]
  5. Citrin CM. Multi-planar reconstruction as a method of eliminating streak artifact in computed tomographic images. Comput Radiol 1982;6:377–378.[CrossRef][Medline]
  6. Calhoun PS, Kuszyk BS, Heath DG, Carley JC, Fishman EK. Three-dimensional volume rendering of spiral CT data: theory and method. RadioGraphics 1999;19:745–764.[Abstract/Free Full Text]
  7. Palestro CJ, Swyer AJ, Kim CK, Goldsmith SJ. Infected knee prosthesis: diagnosis with In-111 leukocyte, Tc-99m sulfur colloid, and Tc-99m MDP imaging. Radiology 1991;179:645–648.[Abstract/Free Full Text]
  8. Palestro CJ, Kim CK, Swyer AJ, Capozzi JD, Solomon RW, Goldsmith SJ. Total-hip arthroplasty: periprosthetic indium-111-labeled leukocyte activity and complementary technetium-99m-sulfur colloid imaging in suspected infection. J Nucl Med 1990;31:1950–1955.[Abstract/Free Full Text]
  9. Stumpe KD, Notzli HP, Zanetti M, et al. FDG PET for differentiation of infection and aseptic loosening in total hip replacements: comparison with conventional radiography and three-phase bone scintigraphy. Radiology 2004;231:333–341.[Abstract/Free Full Text]
  10. White LM, Kim JK, Mehta M, et al. Complications of total hip arthroplasty: MR imaging-initial experience. Radiology 2000;215:254–262.[Abstract/Free Full Text]
  11. Sugimoto H, Hirose I, Miyaoka E, et al. Low-field-strength MR imaging of failed hip arthroplasty: association of femoral periprosthetic signal intensity with radiographic, surgical, and pathologic findings. Radiology 2003;229:718–723.[Abstract/Free Full Text]
  12. Pandey R, Drakoulakis E, Athanasou NA. An assessment of the histological criteria used to diagnose infection in hip revision arthroplasty tissues. J Clin Pathol 1999;52:118–123.[Abstract]
  13. Rockette HE, Gur D, Metz CE. The use of continuous and discrete confidence judgments in receiver operating characteristic studies of diagnostic imaging techniques. Invest Radiol 1992;27:169–172.[CrossRef][Medline]
  14. King JL, Britton CA, Gur D, Rockette HE, Davis PL. On the validity of the continuous and discrete confidence rating scales in receiver operating characteristic studies. Invest Radiol 1993;28:962–963.[CrossRef][Medline]
  15. Dorfman DD, Alf E Jr. Maximum likelihood estimation of parameters of signal detection theory and determination of confidence intervals: rating method data. J Math Psych 1969;6:487–496.[CrossRef]
  16. Dorfman DD, Berbaum KS, Abu-Dagga HI, Schartz KM. RSCORE 4.66. ftp://perception.radiology.uiowa.edu. Accessed November 12, 2004.
  17. Dorfman DD, Berbaum KS, Metz CE. Receiver operating characteristic rating analysis: generalization to the population of readers and patients with the jackknife method. Invest Radiol 1992;27:723–731.[CrossRef][Medline]
  18. Schartz KM, Hillis SL, Berbaum KS, Dorfman DD. MRMC 1.84 BETA 4. Subroutine: analysis of variance with normalized pseudovalue assuming fixed readers. ftp://perception.radiology.uiowa.edu. Accessed November 12, 2004.
  19. Haramati N, Staron RB, Mazel-Sperling K, et al. CT scans through metal scanning technique versus hardware composition. Comput Med Imaging Graph 1994;18:429–434.[CrossRef][Medline]
  20. Wang G, Frei T, Vannier MW. Fast iterative algorithm for metal artifact reduction in x-ray CT. Acad Radiol 2000;7:607–614.[CrossRef][Medline]
  21. Robertson DD, Weiss PJ, Fishman EK, Magid D, Walker PS. Evaluation of CT techniques for reducing artifacts in the presence of metallic orthopedic implants. J Comput Assist Tomogr 1988;12:236–241.[Medline]
  22. Wilting JE, Timmer J. Artefacts in spiral-CT images and their relation to pitch and subject morphology. Eur Radiol 1999;9:316–322.[CrossRef][Medline]
  23. Vannier MW. Current technologies and future directions of musculoskeletal imaging with multi-detector row CT. In: Buckwalter KA, Kransdorf MJ, eds. Syllabus: musculoskeletal imaging—exploring new limits. Oak Brook, Ill: Radiological Society of North America, 2003; 127–132.
  24. Farber JM. CT arthrography and postoperative musculoskeletal imaging with current multichannel CT system. In: Buckwalter KA, Kransdorf MJ, eds. Syllabus: musculoskeletal imaging—exploring new limits. Oak Brook, Ill: Radiological Society of North America, 2003; 119–126.
  25. Link TM, Berning W, Scherf S, et al. CT of metal implants: reduction of artifacts using an extended CT scale technique. J Comput Assist Tomogr 2000;24:165–172.[CrossRef][Medline]
  26. Munk PL, Vellet AD, Levin MF, Romano CC, Lentle B, Bourne RB. Imaging after arthroplasty. Can Assoc Radiol J 1994;45:6–15.[Medline]
  27. Eustace S, Shah B, Mason M. Imaging orthopedic hardware with an emphasis on hip prostheses. Orthop Clin North Am 1998;29:67–84.[CrossRef][Medline]
  28. Vyskocil P, Gerber C, Bamert P. Radiolucent lines and component stability in knee arthroplasty: standard versus fluoroscopically-assisted radiographs. J Bone Joint Surg Br 1999;81:24–26.
  29. Moed BR, Carr SE, Gruson KI, Watson JT, Craig JG. Computed tomographic assessment of fractures of the posterior wall of the acetabulum after operative treatment. J Bone Joint Surg Am 2003; 85-A:512–522.[Abstract/Free Full Text]
  30. Ebraheim NA, Mekhail AO, Darwich M. Open reduction and internal fixation with bone grafting of clavicular nonunion. J Trauma 1997;42:701–704.[Medline]
  31. Slone RM, MacMillan M, Montgomery WJ. Spinal fixation. III. Complications of spinal instrumentation. RadioGraphics 1993;13:797–816.




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ohashi, K.
Right arrow Articles by Berbaum, K. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ohashi, K.
Right arrow Articles by Berbaum, K. S.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE